Focusing on growth

Yesterday’s blog post looked at an infographic that used an area chart with misleading scaling to emphasise the continued increase in total weekly hours worked, a measure of UK employment. Here is a reminder of the published chart:

CityAM infographic

Source: City AM, January 23 2014, p2

So what can we do to improve upon this? The data in the chart is freely available on the website of the Office for National Statistics, so I have done some simple analysis of the data set in Excel. The first chart below plots exactly the same data as the infographic, using a simple line chart instead of the area chart. The scaling has been determined automatically by Excel and the origin is not included.  Notice how the rise in hours worked does not appear quite as pronounced as in the original area chart.

Trend1

 

What happens if we adjust the scaling of the chart to include zero?  The growth in hours worked becomes barely perceptible, and this chart fails to convey any insight:

Trend2

 

The key to choosing the right graph is to focus on the message that we are trying to convey.  In this case, we are probably most interested in showing the year on year growth in weekly hours worked.  So let’s do some work on behalf of the reader and transform the whole data series into percentages showing the growth of each quarter on the corresponding quarter 12 months previously.  We can now comfortably plot the data series as a line chart without having to worry about whether to include the origin or not, and a clear message now stands out from the data.

Chart3

 

The annual growth in total weekly hours worked has hovered around the 2% level for the last 18 months, with the last quarter’s value standing at 2.8%.  It remains to be seen if the last quarter’s numbers represent a significant turning point in employment growth given the relative volatility in the historical data.

Comments

  1. Gerald Davies

    Focus on the message is key, I agree. A good example is what I describe as the asymmetry of growth – if a variable falls by 50% then it needs to double just to get back to square one. Charts often fail to capture the delta.

    Reply

Leave a Reply